Building The Analytics Team At Wish

Wish grew from a successful consumer app to a multi-billion-dollar operation, but its analytics stack collapsed under demand. Early on engineers were the only ones able to pull data, leading to weeks-long turnaround times and constant fire-fighting. The core insight is that a data organization cannot function if it relies on a single skill set; you must hire strong analysts to own reporting while engineers focus on building scalable pipelines. This division of labor prevents both sides from being overloaded and creates a sustainable growth path.

The rebuild started by separating the data pipeline from the reporting layer. The team adopted Luigi for dependency-managed ETL, moved raw data into Redshift and BigQuery, and introduced Looker as a company-wide BI front-end. By providing a dedicated analytics warehouse, query latency dropped dramatically and analysts could generate insights without deep engineering involvement. The shift also lowered the technical barrier for hiring analysts, because the data source became SQL-friendly rather than requiring custom Python scripts over MongoDB.

Prioritization was another critical piece. The team evaluated each request by impact versus effort, fixing bugs first, then tackling high-impact feature requests, and pushing low-value exploratory asks back. This simple matrix allowed a tiny team of analysts and engineers to handle a flood of requests without burning out. It also gave leadership a clear view of where data work delivered the most business value.

The overall lesson for technical leaders is to build a balanced data organization early: hire both engineers and analysts, invest in reliable pipelines and a central warehouse, and enforce a disciplined impact-vs-effort framework for incoming work. Doing so turns analytics from a bottleneck into a strategic asset that scales with the company.